Family Practice Vol. 21, No. 5 © Oxford University Press 2004, all rights reserved.
Better Prescribing Project: a randomized controlled trial of the impact of case-based educational modules and personal prescribing feedback on prescribing for hypertension in primary care
The University of Western Ontario, Faculty of Medicine and Dentistry, Health Sciences Addition, London, Ontario N6A 5C1, Canada
E-mail: carol.herbert{at}fmd.uwo.ca
Received 15 January 2004; Accepted 17 May 2004.
Herbert CP, Wright JM, Maclure M, Wakefield J, Dormuth C, Brett-MacLean P, Legare J and Premi J. Better Prescribing Project: a randomized controlled trial of the impact of case-based educational modules and personal prescribing feedback on prescribing for hypertension in primary care. Family Practice 2004; 21: 575–581.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Objective. The purpose of this study was to assess the impacts of individualized prescribing feedback and interactive small group education aimed at encouraging evidence-based prescribing in family/general practice.
Methods. A two-by-two factorial randomized controlled trial was carried out involving 200 family physicians in British Columbia, Canada. The physicians met monthly in 28 peer learning groups within the Practice-Based Small Group (PBSG) learning programme. Personalized prescribing feedback related to hypertension was provided through ‘prescribing portraits’ which graphically displayed comparative rates of individual and peer group prescribing, together with a concise guide for evidence-based prescribing. A case-based educational module, containing the same evidence-based message, was discussed in small groups. Groups were matched and randomized into four arms of seven groups each: control (n = 56), prescribing portrait only (n = 48), educational module only (n = 47), both portrait and module (n = 49). The main outcome measure was changes in ‘prescribing preferences’ to new patients among those medications used to treat hypertension (i.e. probability that a patient would receive the evidence-based medication as first-line therapy).
Results. Using data from the provincial pharmacy registry (PharmaNet), prescribing preferences for antihypertensive agents were determined for all groups for a 6 month period before and after the interventions, based on 4394 patients receiving a first-line antihypertensive. Significant absolute increases in prescribing preference for thiazides were documented for both the module +0.068 (confidence interval [CI] 0.022–0.115) and the portrait +0.065 (CI 0.018–0.111). Preference in the group receiving both module and portrait increased by +0.115 (CI 0.040–0.189).
Conclusions. Evidence-based educational interventions combining personalized prescribing feedback with interactive group discussion can lead to modest but meaningful changes in physician prescribing. Clear messages, proper trial design and sensitive outcomes are necessary to demonstrate these changes.
Keywords. Continuing medical education, general practice, prescribing feedback, problem-based learning, physician behaviour change.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Although large sums of money are spent on interventions to improve physician performance, most studies show only moderate impacts at best.1,2 Some studies of audit and feedback have demonstrated impacts that are potentially worthwhile.2,3
Individualized feedback with specific recommendations, especially when combined with education, generally have been more effective than single interventions.4–6 However, there are very few studies of educational interventions to improve prescribing in primary care, and the need for better randomized controlled studies has been stressed.7,8
Three recent randomized controlled trials specifically aimed at evidence-based prescribing of antibiotics in primary care have shown encouraging results. In Canada, a study of mailed prescribing feedback, accompanied by educational bulletins, was associated with a 2.6% increase in first-line antibiotic prescribing among physicians who received the intervention, while first-line prescribing decreased by 1.7% in the control group.9 However, similar trials have reported no positive impact when the prescribing feedback was unsolicited.10,11 A Swedish study using prescription feedback, coupled with small group case-based discussion, demonstrated improvements in prescribing of first-line antibiotics for urinary tract infection.12 A similar study in Norway showed an increase in evidence-based prescribing ranging from 6 to 13%.13
A recent study of prescribing for a variety of conditions in general practice in New Zealand, using educational interventions in clinical practice groups, showed that changes were greatest for short-term prescriptions such as antibiotics.14
The Better Prescribing Project was designed to assess the impact of (i) individualized feedback through ‘prescribing portraits’ (histograms comparing an individual's prescribing rates with those of the physician's small group and of all physicians in the study), combined with a concise evidence-based statement to guide future prescribing; and/or (ii) evidence-based case-based educational modules discussed in existing learning groups. ‘Evidence-based’ here means prescribing that is supported by recent systematic reviews of current evidence from randomized controlled trials. This paper presents the results from an intervention designed to affect prescribing for hypertension.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Design
The study used a factorial design with randomization of matched groups of physicians to four arms (Fig. 1): both experimental educational module and prescribing portrait (EP); control educational module and portrait (CP); experimental educational module only (EN); and control module only (CN). The details of the process of matching and random allocation are provided in Box 1.
|
| BOX 1 Randomization process Physician groups first were paired by approximate group size, sex ratio and average length of participation in the PBSG programme. A random number was assigned to each pair—allocating one group to ‘module: heads’ and the other group to ‘module: tails’. Each pair then was matched with another pair by the same three variables, making a matched ‘quartet’ of four groups. Another random number was assigned that allocated the first pair of groups to ‘portrait: heads’ and the second pair to ‘portrait: tails’. Finally, to produce better overall balance of baseline characteristics between the treatment and control groups than could be achieved by pair matching alone, the above random numbers were re-assigned until participants in the four arms of the trial differed by <10% on the following variables: year of graduation (a proxy for age of physician), patients per week (a proxy for size of practice), solo practice, number of other physicians in a group practice and percentage of participants who see drug representatives at least once a month. Blind to ‘heads’ and ‘tails’ assignment, the principal investigator then publicly tossed a coin twice for each quartet—allocating each PBSG within the quartet to one of the four experimental arms: (i) control (CN); (ii) portrait only (CP); (iii) module only (EN); or (iv) both portrait and module (EP).
|
Participants
British Columbia has 
6000 actively prescribing physicians, of whom 4000 are general family practitioners. Similar to other provinces in Canada, 10–15% of these family physicians participate in peer learning groups as part of the Practice-Based Small Group (PBSG) learning programme15 organized by the Foundation for Medical Practice Education (FMPE), based at McMaster University (www.fmpe.org).
Letters of invitation were sent to all 60 PBSG group in British Columbia in 1999. The letter offered participants personal prescribing feedback for several ‘common and important medical problems encountered in general practice’. Inclusion criteria for groups were: at least three and no fewer than 50% of members agreed to participate; participants were in active practice; and participants consented to examination of their prescribing data as captured by PharmaNet. PharmaNet is an on-line central pharmacy database that records all prescriptions dispensed by community pharmacies in British Columbia.
Two hundred physicians in 28 PBSGs joined the trial (Fig. 1), with an additional PBSG serving an advisory role to pre-test the materials. A further seven physicians in two groups joined the trial subsequent to the hypertension intervention, and are therefore not included in this report. Groups that declined stated that this was either because they were unable to meet the perceived research time commitment or because the continuation of the group throughout the entire study period was in doubt.
The research was approved by the University of British Columbia Behavioural Research Ethics Committee and informed consent was obtained from all participants.
Interventions
Between May and September 1999, the experimental educational module and prescribing portraits, designed to promote increased use of thiazide diuretics as first-line treatment for hypertension, were used by the intervention groups during their regular PBSG meetings.
Each 1-page prescribing portrait contained (i) histograms comparing an individual's prescribing rates for antihypertensives with those of the physician's small group and of all physicians in the study; and (ii) a succinct evidence-based message to guide future prescribing. On the back of the page was an explanation of how the histogram was derived, plus potential limitations of the data. The format and content of these portraits were pre-tested for clarity and accuracy by the advisory PBSG.
Following usual PBSG guidelines and process, an evidence-based educational module was developed by a team of British Columbia family physicians and specialists, with review by content specialists. The 8 page module presents three representative patient cases, together with an information section that presents the evidence to guide management. The module was discussed in small groups during a single session of <2 h, and participants were encouraged to decide on management of each case. Commentaries are included at the end of the module-describing management of the cases in relation to the evidence.
Normally, at the end of a PBSG session, each group facilitator completes a brief form that documents the group's assessment of the session, areas for follow-up and the predicted impact, if any, on their practice. This form was modified slightly to record additional group process data, and all participants who received the experimental modules completed this modified form individually. Participants who received portraits individually completed a brief questionnaire concerning their understanding of the portraits and any suggestions for improvement.
The controls for the experimental module received an unrelated module of their choice produced by FMPE on topics such as cancer pain or sinusitis. The controls for the prescribing portraits received no portraits. The project office was the sole distributor of experimental materials in British Columbia during the course of the study, ensuring that physicians in the control group did not access them.
Outcome impact measure—physician prescribing preference
The overall hypothesis was that the prescribing portrait plus the experimental module would influence physicians' preference for target drugs when prescribing to patients newly initiated on drug therapy for hypertension. Clinically significant improvement would be a change of 10%. Preference was defined as the probability of choosing a particular antihypertensive agent, given that the patient was prescribed any antihypertensive for the first time.15 In other words, a doctor's ‘preference for thiazides’ was the fraction of that doctor's patients who received a thiazide (numerator) among all of his or her patients who received antihypertensives (denominator) for the first time in 365 days. An operational definition for newly treated hypertensive patients was developed (Box 2).
| BOX 2 Operational definition for diagnosis of hypertension The operational definition of newly treated hypertension was: all patients who had received at least one antihypertensive drug for the first time in at least 365 days, but who did not receive furosemide (a marker of congestive heart failure) or a nitrate (a marker of angina). This number was the denominator used to calculate the ‘prescribing preferences’ for antihypertensives.
|
Data analysis
Prescription records for participating physicians for all drugs in the PharmaNet database relevant to the study were analysed. Patients' and physicians' unique identification numbers were encrypted to preserve anonymity. To determine changes, we measured prescribing preferences from the 6 month period prior to mailing of the materials and the combined 1–6 month period after the intervention. The difference in preference after the intervention versus before the intervention was termed the ‘preference difference’ (PD). To adjust for any prescribing changes in the control group, and thus provide a more accurate measure of the impacts of the interventions, we subtracted the control group's PD from the intervention group's PD. This difference between PDs, was termed the ‘adjusted preference difference’ (APD). To obtain confidence intervals (CIs), identical tables were constructed for the same physicians and same dates in 1998, and the variance among groups versus within groups was measured. These estimates of variance were used to adjust the variance for clustering of physicians within small groups, and for patients within physicians.
Analysis of impacts on prescribing was blinded by labelling the four parallel arms with arbitrary codes, and the investigators were blinded to the intervention status until the analysis of prescribing impact was complete. To assess need to adjust for clustering of effects, a C-statistic was calculated and found to be less than zero—indicating no clustering.17
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
At baseline (i.e. in the 6 months prior to the intervention), among 2232 patients initiated on antihypertensive therapy, the overall preference for prescribing thiazide diuretics was 0.31. In other words, when participating physicians decided to prescribe a blood pressure-lowering medication to a patient who had not used one in the previous year, the probability was 31% that the patient would receive a thiazide diuretic as first-line therapy.
Receiving the module increased preference for thiazides by 6.8% (APD = 0.068; 95% CI 0.022–0.115) in the 6 months following the intervention. At the same time, the portrait was associated with a 6.5% increased preference for thiazides (APD = 0.065; 95% CI 0.018–0.111). These changes in prescribing were the same at 4–6 months as at 1–3 months after the intervention (Fig. 2). The group that received both the module and the portrait had the greatest increase in preference for thiazides (APD = 0.115; 95% CI 0.040–0.189) (Fig. 3).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Principal findings
This study demonstrated that educational interventions with clear messages can lead to meaningful changes in physician prescribing behaviour. In this trial, both interventions were associated with significant absolute increases in the use of thiazides as first-line therapy for hypertension—modules (6.8%) and portraits (6.5%). The combined intervention, consisting of both the educational module and portrait, had the strongest impact (11.5%).
Strengths and weaknesses, important differences in results
The outcome measure, preference for first-line therapy, had much greater specificity as a measure of physicians' decision making than other commonly used measures that are contaminated with prescription renewals. Although participants clearly could not be blind to the intervention status, investigators were blind until the end of data analysis. Participant feedback showed a high level of interest in the process of the study and that the interventions (hypertension module and portrait) were well accepted.
The change in overall prescribing rates documented here are modest, in keeping with similar studies.12,13 It is noteworthy that a single intervention combining prescribing feedback and a brief interactive educational session, with no follow-up prescribing feedback, would be sufficient to effect this degree of change in the prescribing practices of participating physicians.4,5,18 These results also compare favourably with the increase in guideline-based prescribing (5.2%) found in a recent randomized controlled trial of educational outreach in general practice.19
Meaning of the study and implications for clinicians and policy makers
Although the changes are modest, the impacts of improved prescribing for hypertension are significant, both for patients and for the health care system. If the changes in prescribing that occurred in the experimental groups in British Columbia were reproduced in the >3000 family physicians in Canada registered in the PBSG learning programme, the increased first-line use of thiazides in treatment of hypertension would have substantial cost-saving implications for the health system—especially when approximate thiazide daily cost ($0.01) is compared with approximate calcium channel blocker daily cost ($1.00–3.00). These savings would apply to the 50% of patients who will require only one drug to control their hypertension and would be sustained as long as they stay on the first-line drug. Fretheim et al.20 have shown convincingly the potential savings with very conservative changes in prescribing in six countries, including Canada.
Can these results be generalized? Whether similar results could be obtained for physicians not already registered in a similar continuing medical education (CME) programme is unknown. Although physicians who voluntarily participate in ongoing CME may be different from non-participants, analysis of the overall prescribing patterns for three clinical conditions (hypertension, type 2 diabetes and congestive heart failure) did not identify any significant differences between PBSG participants and other family/general practitioners in British Columbia. Similarly, although it might be assumed that this group of physicians would be more open to educational or feedback interventions, the rates of intention to change statements reported by participants in this study21 are actually lower than those reported for other CME events.22,23
Unanswered questions and future research
Changing the clinical practice of physicians is known to be a very complex process, with interwoven factors and formidable barriers.24,25 This study showed that the combination of prescribing feedback and interactive education produced prescribing change, but other similar studies have not. Many questions remain about how construction of the interventions and the selection of participants influence outcomes. Additional rigorous studies of these combined interventions, that modify other key characteristics but maintain sensitive outcome measures, are necessary to understand better the factors in these changes.
|
Declaration |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Funding: This study has been made possible by a financial contribution from the Health Transition Fund, Health Canada.
Ethical Approval: University of British Columbia
Conflicts of interest: None.
|
Acknowledgments |
|---|
Maud Van Breemen created the personal prescribing profiles. Laura Esmail provided exceptional support as research assistant. The secretarial assistance of Nancy Fedyk is gratefully acknowledged. The views expressed herein do not necessarily represent the official policy of Health Canada.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
1 Freemantle N, Harvey EL, Wolf F, Grimshaw JM, Grilli R, Bero LA. Printed educational materials: effects on professional practice and health care outcomes (Cochrane Review). In The Cochrane Library, Issue 3. Oxford: Update Software; 2000.
2 Davis DA, Thomson MA, Oxman AD, Haynes RB. Changing physician performance: a systematic review of the effect of continuing medical education strategies. J Am Med Assoc 1995; 274: 700–705.[Abstract]
3 Thomson O'Brien MA, Oxman AD, Davis DA, Haynes RB, Freemantle N, Harvey EL. Audit and feedback: effects on professional practice and health care outcomes. Cochrane Database of Sytematic Reviews Issue 2; 2001.
4 Thomson O'Brien MA, Oxman AD, Davis DA, Haynes RB, Freemantle N, Harvey EL. Audit and feedback versus alternative strategies: effects on professional practice and health care outcomes. Cochrane Database of Systematic Reviews Issue 2; 2001.
5 Gill PS, Makela M, Vermeulen KM et al. Changing doctor prescribing behaviour: a review. 1998. In Effective Health Care: Getting Evidence into Practice. York: NHS Centre for Reviews and Dissemination, The University of York; 1999; 5: 1–16.
6 Wensing M, Van der Weijden T. Implementing guidelines and innovations in general practice: which interventions are effective? Br J Gen Pract 1998; 48: 991–997.[ISI][Medline]
7 Grimshaw JM, Shirran L, Thomas R et al. Changing provider behavior: an overview of systematic reviews of interventions. Med Care 2001; 39 (8 Suppl 2): II2–II45.[ISI][Medline]
8 Figuerias A, Sastre I, Gestal–Otero JJ. Effectiveness of educational interventions on the improvement of drug prescription in primary care: a critical literature review. J Eval Clin Pract 2001; 7: 223–241.[CrossRef][ISI][Medline]
9 Hux JE, Melady MP, DeBoer D. Confidential prescriber feedback and education to improve antibiotic use in primary care: a controlled trial. Can Med Assoc J 1999; 161: 388–392.
10 O'Connell DL, Henry D, Tomlins R. Randomised controlled trial of effect of feedback on general practitioners' prescribing in Australia. Br Med J 1999; 318: 507–511.
11 Sondergaard J, Andersen M, Vach K, Kragstrup J, Maclure M, Gram LF. Detailed postal feedback about prescribing to asthma patient combined with a guideline statement showed no impact: a randomised controlled trial. Eur J Clin Pharmacol 2002; 58: 127–132.[CrossRef][ISI][Medline]
12 Lundborg CS, Wahlstrom R, Oke T, Thomson G, Diwan V. Influencing prescribing for urinary tract infection and asthma in primary care in Sweden: a randomized controlled trial of an interactive educational intervention. J Clin Epidemiol 1999; 52: 801–812.[CrossRef][ISI][Medline]
13 Lagerlov P, Loeb M, Andrew M, Hjortdahl P. Improving doctors' prescribing behaviour through reflection on guidelines and prescription feedback: a randomised controlled study. Qual Health Care 2000; 9: 159–165.
14 Richards D, Toop L, Graham P. Do clinical practice education groups result in sustained change in GP prescribing? Fam Pract 2003; 20: 199–206.
15 Premi J, Shannon S, Hartwick K, Lamb S, Wakefield J, Williams J. Practice-based small-group CME. Acad Med 1994; 69: 800–802.[ISI][Medline]
16 Maclure M, Dormuth C, Naumann T et al. Influences of educational interventions and adverse news about calcium-channel blockers on first-line prescribing of antihypertensive drugs to elderly people in British Columbia. Lancet 1998; 352: 943–948.[CrossRef][ISI][Medline]
17 Donner A, Klar N. Methods for comparing event rates in intervention studies when the unit of allocation is a cluster. Am J Epidemiol 1994; 140: 279–289.
18 Davis D, O'Brien MAT, Freemantle N, Wolf FM, Mazmanian P, Taylor-Vaisey A. Impact of formal continuing medical education: do conferences, workshops, rounds and other continuing education activities change physician behavior or health outcomes? J Am Med Assoc 1999; 282: 867–874.
19 Freemantle N, Nazareth I, Eccles M, Wood J, Haines A. A randomised controlled trial of the effect of educational outreach by community pharmacists on prescribing in UK general practice. Br J Gen Pract 2002; 52: 290–295.[ISI][Medline]
20 Fretheim A, Aaserud M, Oxman AD. The potential savings of using thiazides as the first-choice antihypertensive drug: cost-minimisation analysis. BMC Health Serv Res 2003; 3: 18.[CrossRef][Medline]
21 Wakefield J, Herbert CP, Maclure M et al. Commitment to change statements can predict actual change in practice. J Contin Educ Health Prof 2003; 23: 81–93.[CrossRef][Medline]
22 Fidler H, Lockyer JM, Toews J, Violato C. Changing physicians' practices: the effect of individual feedback. Acad Med 1999; 74: 702–714.[ISI][Medline]
23 Mazmanian PE, Ratcliff-Daffron S, Johnson RE, Davis DA, Kantrowitz MP. Information about barriers to planned change: a randomized controlled trial involving continuing medical education lectures and commitment to change. Acad Med 1998; 73: 882–886.[ISI][Medline]
24 Lockyer JM, Fidler H, Ward R, Basson RJ, Elliott S, Toews J. Commitment to change statements: a way of understanding how participants use information and skills taught in an educational session. J Contin Educ Health Prof 2001; 21: 82–89.[CrossRef][Medline]
25 Mazmanian PE, Davis D. Continuing medical education and the physician as a learner. J Am Med Assoc 2002; 288: 1057–1060.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  L. Cranston Funding for continuing medical education Can. Med. Assoc. J., June 3, 2008; 178(12): 1578 - 1578. [Full Text] [PDF]
|
|
|
|
 |
|
 H. Armson, S. Kinzie, D. Hawes, S. Roder, J. Wakefield, and T. Elmslie Translating learning into practice: Lessons from the practice-based small group learning program Can Fam Physician, September 1, 2007; 53(9): 1477 - 1485. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. R Norman and K. W Eva Does clinical experience make up for failure to keep up to date? Evid. Based Med., June 1, 2005; 10(3): 66 - 68. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||